DAPK2 Downregulation Associates With Attenuated Adipocyte Autophagic Clearance in Human Obesity

Adipose tissue function includes the deposition of nutrient-derived energy excess as lipids and the integration of energy balance signals to produce leptin and adiponectin required for metabolic homeostasis. Chronic energy balance disruption as seen in obesity leads to adipose tissue lipid overload and adaptive stress responses toward ultimate dysfunction and metabolic inflexibility (1). Large-scale transcriptomic studies of adipose tissue from obese patients or obese rodent models pointed to inflammatory pathways as a prominent response to lipid overload (2–4). Inflammation in obesity results from tissue remodeling by immune cell infiltration/proliferation within the adipose fat depot (5). Indeed, whereas the adipocyte is the predominant cell type in healthy lean adipose tissue, immune cells can equal or even exceed fat cell number in obesity. Metabolic inflammation is now recognized as a key factor in obesity-related insulin resistance and the subject of current investigations of immune cell subpopulations (6) toward novel therapeutic approaches.

Besides identification of expression networks and associated biological functions, transcriptomic approaches also bring attention to single genes unrelated to annotated pathways. In the current study, we focus on death-associated protein kinase 2 (DAPK2), also called DRP-1, which hits among the most downregulated genes of the adipose tissue transcriptome in human morbid obesity (7). DAPK2 belongs to a family of serine/threonine calmodulin-regulated protein kinases that consists of five members, all displaying high sequence homology in their N-terminal kinase domains, whereas COOH-terminal regions highly diverge. DAPK1, the founding family member, is involved in apoptosis (8) and is suppressed in many tumors (9). Unlike DAPK1, DAPK2 and other DAPKs are devoid of a COOH-terminal death domain, ankyrin repeats, and cytoskeleton-interacting region (9). Therefore, DAPK2 is a small protein that only contains a kinase domain, a calmodulin binding region, and a short tail with undefined structural motifs. DAPK2 is ubiquitously expressed, but its physiological functions are still largely unknown. Interestingly, large-scale RNA interference kinome screens identified DAPK2, suggesting a role in autophagy (10), secretory pathway (11), or transforming growth factor-β signaling by protein-to-protein interaction (12).

In the current study, we investigated the significance of DAPK2 downregulation in human obesity. We first identified adipocytes and not other nonlipid filled adipose tissue cells as targets of DAPK2 downregulation. Then, the effects of DAPK2 gain/loss of function in cultured adipose cells revealed a regulatory role of the kinase in lysosome-mediated remodeling, which led us to further investigate adipocyte autophagic clearance in human obesity and weight loss. We demonstrated attenuation of adipocyte autophagic activity in human obesity, partially reversible after weight loss after bariatric surgery. Thus, these results link DAPK2 downregulation and defective autophagy to fat cell dysfunction in human obesity.

All subjects gave their informed consent in an accepted protocol related to the physiopathology of obesity (Assistance Publique-Hôpitaux de Paris, Clinical Research Contract). Clinical investigations were performed according to the Declaration of Helsinki and approved by local ethics committees. Middle-aged obese subjects who were candidates for bariatric surgery were not on a restrictive diet and their weights were reported stable. All subjects met usual eligibility criteria for bariatric surgery. Follow-up of patients after gastric bypass included monitoring of weight loss and obtaining subcutaneous adipose tissue biopsy specimens from the abdominal region. All tissue samples were processed immediately for adipocyte isolation or frozen for mRNA extraction.

Control subjects were nonobese subjects undergoing plastic surgery or planned surgery for gall bladder ablation or hernia. None had diabetes or metabolic disorders or were taking medication. Main characteristics of the patient groups are summarized in Table 1.

Floating adipocytes were obtained by collagenase digestion essentially as described by Rodbell (13), except that DMEM was used instead of Krebs-Ringer buffer. Floating adipocytes were separated from the stromal vascular fraction (SVF) cells by aspirating the infranatant and rinsing three times with DMEM. Minor contamination of the adipocyte fraction by stromal vascular cells was checked on morphological images of adipocyte suspensions used for fat cell size determination.

3T3-L1 cells were maintained and differentiated in standard conditions. Adenovirus was obtained by recombination of viral backbones with shuttle vectors containing DAPK2 cDNAs and packaging in complementing 293 cells, as described previously (14). GFP-LC3 3T3-L1 stable transfectants were obtained as described (15). Human preadipocytes were isolated from subcutaneous adipose tissue and cultured as described (16). Small interfering (si)RNA–mediated knockdown was performed with 20 nmol/L siRNAs using lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instruction. DAPK2 siRNA sequence was from Abgent (5′-UGUCUGGAGGAGAGCUCUUTTAAGAGCUCUCCUCCAGACATT-3′). Scrambled siRNAs were used as controls. Fresh medium supplemented with antibiotics and 10% SVF was added 16 h after transfection, and cells were incubated for a further 24 h.

Male ob/ob and ob/+ mice were maintained on a normal chow diet with ad libitum feeding and killed at 12–16 weeks of age (body weights 30.1 ± 3.1 vs. 58.2 ± 3.4 g). C57BL/6J male mice were fed a high-fat diet (60% calories from fat) for 16 weeks from weaning onward and killed (body weights 37.9 ± 1.8 vs. 48.2 ± 3.5 g).

mRNAs were reversed transcribed and cDNA were quantified by real-time RT-PCR using qPCR MasterMix Plus for SYBR Green (Eurogentec). The PCR efficiency of primer pairs was checked in standard curves, and expression data were expressed using the ∆∆Ct method. 18S RNA was used for normalization. The sequence of primer pairs for DAPKs was as follows:

• Human DAPK2: 5′-ACGTGGTGCTCATCCTTGA-3′ and 5′-TGGCCTCCTCCTCACTCA-3′.

• Mouse DAPK2: 5′-GACGTGGTGCTCATCCTTG-3′ and 5′-GGCTTCCTCCTCACTTAACGA- 3′.

• Mouse DAPK1: 5′-CCTGATTTCCAGGACAAGG-3′ and 5′-CTTTAGCCACGGAGTAATCAGCC-3′.

• Mouse DAPK3: 5′-ATTTGTACCGGAGGTTCTCG-3′ and 5′- TCTGAAGGATTCTGGGGACA-3′.

A complete list of the other primer sequences is provided in the Supplementary Data.

Experimental design followed state of the art guidelines (17). Freshly isolated floating adipocytes or differentiated cells lines were rinsed with high-glucose DMEM and incubated for 2 h in the presence of 100 nmol/L bafilomycin A1, 100 µmol/L leupeptin/20 mmol/L NH₄Cl, or 25 µmol/L chloroquine at 37°C in a humidified 95% air/5% CO₂ atmosphere. Cells were collected in lysis buffer (50 mmol/L Tris [pH 7.4], 0.27 mol/L sucrose, 1 mmol/L Na-orthovanadate, 1 mmol/L EDTA, 1 mmol/L EGTA, 10 mmol/L Na β-glycerophosphate, 50 mmol/L NaF, 5 mmol/L Na pyrophosphate, 1% [w/v] Triton X-100, and 0.1% [v/v] 2-mercapto ethanol) supplemented with Complete protease inhibitors and stored frozen at −20°C before further processing.

Cell lysates were processed as previously described (18). Commercial antibodies against LC3, Akt, and β-actin were from Cell Signaling Technology, DAPK2 from ProSci, and p62 from Progen.

Lipolytic rates were assessed in cells maintained for 2 h in phenol-red free medium containing 2% BSA with or without 10⁻⁵ mol/L isoproterenol for lipolytic stimulation. Glycerol release was measured using a glycerol colorimetric assay kit (Cayman Chemical, Ann Arbor, MI) and normalized to cell protein content.

3T3-L1 stable transfectants expressing GFP-LC3 were differentiated and processed for fluorescent imaging as previously described (19). Intracellular lipid droplets and floating adipocytes were sized using Perfect Image (Clara Vision) from phase-contrast images. Immunofluorescence on paraffin-embedded adipose tissue sections was performed as described (20).

Results are expressed as mean ± SEM. The Mann-Whitney U test was used in all analyses except for paired data, which were tested with the Wilcoxon signed rank test or the paired Student t test. The Spearman rank correlation test was used for correlations.

Our attention on DAPK2 was brought by pangenomic transcriptome analysis comparing adipose tissue gene expression in lean and obese subjects (7). We confirmed a threefold decrease in DAPK2 mRNA (Fig. 1A) in subcutaneous adipose tissue samples from 65 severely obese (BMI range 34–79 kg/m²) compared with 10 nonobese subjects (BMI range 20–23 kg/m²), not influenced by type 2 diabetes (Fig. 1B) or by sex (not shown). Low DAPK2 mRNA expression was also independently confirmed in a subgroup of 24 severely obese compared with 10 overweight (BMI 24–28 kg/m²) subjects (Fig. 1C), pointing to DAPK2 downregulation as a feature of severe obesity but not of simple overweight. Relative DAPK2 mRNA levels in obese subjects positively correlated with HDL cholesterol and inversely associated with triglyceridemia and serum interleukin 6 (IL-6). A trend for negative association to subcutaneous fat cell size (P = 0.057) was also observed (Table 2).

Because bariatric surgery (i.e., gastric bypass) is the most efficient intervention to induce weight loss in morbid obesity, we examined DAPK2 mRNA in 10 patients over time at 0, 3, 6, and 12 months after surgery. Gradual time-dependent recovery, almost complete after 12 months (Fig. 1D), indicated that DAPK2 loss was not irreversible in obese adipose tissue. We also found decreased DAPK2 mRNA in the subcutaneous adipose tissue of mice with diet-induced obesity, indicating common obesity-related DAPK2 regulation in humans and rodents (Fig. 1E). Adipose tissue digestion by collagenase was performed to determine relative DAPK2 mRNA expression in the isolated adipocyte and the nonlipid-filled stromal vascular cell compartment of human fat tissue. In nonobese as well as in obese subjects, DAPK2 mRNA was more abundant in adipocytes than in stromal vascular cells (Fig. 1F), and a major reduction with obesity was found in the adipocyte fraction (Fig. 1F). Thus, adipocyte-enriched DAPK2 expression is suppressed in human obesity and reversibly recovers after bariatric surgery–induced weight loss in humans. Immunofluorescence labeling of adipose tissue sections confirmed adipocyte DAPK2 suppression in obese patients at the protein level (Fig. 1G).

The bulk of human fat cells are renewed every 10 years, consistent with function in long-term energy storage, and the adipocyte is considered a long-lived cell type (21). In this context, the conserved process of intracellular organelle remodeling by lysosomes (i.e., autophagy) is crucial to maintain homeostasis (22), especially in oxidatively prone conditions linked to a lipid-rich adipocyte environment. Because DAPK2 was suggested as a potential autophagy regulator (10,23,24), we focused on constitutive autophagy in the presence of nutrients. We generated 3T3-L1 adipocytes stably expressing GFP-LC3, which retained full differentiation capabilities, as judged by the presence of large multilocular cytoplasmic lipid droplets. Fluorescent GFP-LC3 distribution identified autophagosome vesicles, visualized as punctuated labeled structures (Fig. 2A) and also detectable as LC3-II fast-migrating bands on Western blots (Fig. 2B), which accumulated in the presence of lysosomal inhibitors such as chloroquine, leupeptin/NH₄Cl, or bafilomycin A1, indicating ongoing autophagic clearance. As expected, nutrient withdrawal from culture medium increased autophagy flux (Fig. 2C) in agreement with well-known induction of autophagy by cell starvation. In some cell types, such as hepatocytes (25) or macrophages (26), intracellular lipid droplet organelles can end up in lysosomes for lipid degradation through so-called lipophagy. Lipophagy is poorly documented in fat cells, but the importance of activation of cytoplasmic neutral lipases for lipid droplet degradation is clearly established (27). We observed no increase in autophagic flux during acute lipolytic stimulation of 3T3-L1 adipocytes with the β-adrenergic agonist isoproterenol (Fig. 2D), and lysosomal inhibitors only marginally affected lipid mobilization as judged by cell glycerol release into the medium (Fig. 2E). Thus, autophagic clearance and lipid store mobilization by catecholamines are distinct processes in adipocytes.

DAPK2 mRNA expression in terminally differentiated 3T3-L1 adipocytes was two orders of magnitude lower than the two other related DAPKs (Fig. 3A), pointing to the 3T3-L1 model as a suitable system to explore the effect of exogenous DAPK2. Adenoviral (Ad)-based expression of DAPK2 (Ad-DAPK2) or a kinase-dead mutant version (Ad-DAPK2-K52A), coexpressed with an EGFP reporter, indicated a high proportion of transduced fat cells after 24 h (not shown). Compared with cells transduced with a null virus, DAPK2 expression did not produce overt metabolic effects. Cells continued to respond to acute insulin stimulation by massive Akt phosphorylation (Fig. 3B) and still contained lipid droplets with normal size distribution (Fig. 3C). Metabolic (leptin, adiponectin, fatty acid synthase, or perilipin 1) and inflammatory (IL-6, chemokine [C-C motif] ligand-2) gene expression was also unaltered (Fig. 3D). Furthermore, we observed no sign of cell leakage by cytoplasmic lactate dehydrogenase release (Fig. 3E), indicating unaffected cell viability. Considering a trend (+20%) toward increased ability to release glycerol upon adrenergic lipolytic stimulation (Fig. 3F), DAPK2-expressing adipocytes showed no evidence for dysfunction or death. Adipocyte clearance by autophagy (i.e., autophagic flux), as measured by LC3-II accumulation in the presence of lysosomal inhibitors, was higher in cells expressing DAPK2 than in cells transduced with viral backbone alone (Fig. 3G). Furthermore, DAPK2-mediated stimulation of autophagic clearance was abolished with a kinase-dead mutant version, indicating dependence on kinase activity for autophagy modulation (Fig. 3H).

We next evaluated the effect of DAPK2 inhibition by siRNA in human preadipocytes obtained from adipose tissue SVF. siRNA sequences decreased DAPK2 protein by half (Fig. 4A), increased cell contents of the p62 autophagic substrate (Fig. 4B), and decreased autophagosome accumulation with lysosome inhibitors, indicating reduced autophagic flux (Fig. 4C). All together, these data on gain or loss of function demonstrate that modulation of DAPK2 expression in adipose cells associates with basal autophagic tone regulation.

The above results suggest that downregulated adipocyte DAPK2 might be linked to attenuation of autophagic clearance in obesity. We therefore prepared isolated adipocyte cell fractions from obese and lean subcutaneous fat biopsy specimens and evaluated steady-state levels of the autophagic substrate p62, which were elevated twofold in obese versus lean adipocytes (Fig. 5A and B), indicative of lower autophagic degradation. Nevertheless, p62 mRNA was also higher in adipocytes from obese subjects compared with control subjects (Fig. 5C), precluding any definite conclusion on autophagic flux based on p62 protein content. We next set up an experiment in which freshly isolated adipocytes were incubated in a standard nutrient-rich medium, in the presence or absence of lysosome inhibitors, to evaluate rates of autophagosome accumulation (i.e., adipocyte autophagic flux). LC3-II accumulation in the presence of lysosome inhibitors was obvious in nonobese adipocytes but was only barely detected in obese fat cells (Fig. 5D). Quantitative analysis of LC3-II accumulation rates demonstrated reduced adipocyte autophagic clearance in obesity (Fig. 5E). Furthermore, adipocyte autophagic flux inversely correlated with fat cell size (Fig. 5F), which is primarily determined by lipid amounts within the unilocular lipid droplet. Thus, attenuated autophagic flux in obesity is linked with adipocyte lipid overload. Similarly, adipocytes isolated from obese (ob/ob) mice also exhibited lower autophagic activity than nonobese (ob/+) controls (Fig. 5G and H), indicating common reduction in obese rodents and humans.

Reversibility of autophagy attenuation was investigated by assessing adipocyte preparations obtained from obese patients after bariatric surgery–induced weight loss. In nine patients, adipose tissue samples were obtained pre- and postsurgery at one occasion within 3 to 12 months after the intervention. All patients lost weight and reduced subcutaneous adipocyte cell size and lipid contents (Fig. 6A), although at different degrees, because of the large time frame in postsurgery sample collection. Interestingly, total adipocyte LC3 protein content pre- versus postsurgery changed proportionally to the extent of adipocyte size reduction (Fig. 6B). Pre- versus postsurgery comparisons indicated ameliorated adipocyte autophagic clearance in all patients (Fig. 6C) and significant recovery, regardless of the use of chloroquine or leupeptin to evaluate autophagic flux (Fig. 6D and E). Thus, adipocyte lysosome–mediated remodeling is compromised in obesity, and partial recovery can be obtained by weight loss.

Increasing evidence points to the critical role of autophagy in metabolic diseases, linked to the adaptive response to chronic metabolic stress (28). In almost every tissue participating in energy homeostasis (liver, pancreas, muscle, and even hypothalamus), invalidation of autophagy by tissue-specific gene knockout was found to be associated with metabolic dysfunction (25,29–33). However, adipocyte responses are still poorly understood because invalidation of adenine thymine guanine alters normal fat tissue differentiation (34,35). Our study brings new evidence that autophagy attenuation associates with adipocyte dysfunction in obesity. Interestingly, previous reports showed autophagy activation in adipose tissue in human or mice obesity (36–38). Noteworthy, the study of total adipose tissue takes account of obesity-associated inflammatory cells, in which autophagy is linked to immune function, so that it remains unclear from these reports in which cell type autophagy is modulated. Here, specific focus on adipocytes and direct measurement of autophagosome clearance show that obesity is associated with downregulation of adipocyte autophagic turnover, partially recovering after bypass surgery–mediated weight loss. Our data are in line with a negative role of lipids on autophagosome dynamics (39,40) and with high-fat diet–induced autophagy defects previously demonstrated in the liver (41,42), kidney (43), and hypothalamus (44).

Autophagic degradation of lipid droplets, also called lipophagy, can decrease the intracellular lipid burden in many cell types but is shown here to only marginally modulate in vitro adipocyte lipolysis. This is in good agreement with the prominent role of neutral cytoplasmic lipases in fat cell lipid mobilization (27) and suggests that autophagy might control other adipocyte phenotypes. Indeed, autophagy inhibition potently induces cell inflammation (45), including fat cell inflammation (37,46). Furthermore, mice with systemic haploinsufficiency for the Atg7 autophagy gene overproduce reactive oxygen species in adipose tissue when obese (47). Thus, we favor the possibility that autophagy serves in the long-term to dampen adipocyte inflammation linked to metabolic dysfunctions.

The current study links obesity-related downregulation of a kinase, DAPK2, with attenuation of adipocyte autophagy. DAPK2 was reported proapoptotic when transiently overexpressed (48,49). In the context of adipose cells lines with low endogenous expression, we found no evidence for cell death induction after Ad-mediated DAPK2 protein expression. Furthermore, by retroviral-mediated gene transfer, we could obtain viable 3T3-L1 adipocyte clones stably expressing wild-type DAPK2 but not the constitutively active form. Thus, we believe that DAPK2 is not proapoptotic in the adipocyte cell environment, likely because of fine-tuning of kinase activation by appropriate mechanisms. Our present data, rather, indicate that DAPK2 is linked to constitutive autophagy and provide evidence that kinase activity is required for this regulation. Molecular mechanisms by which DAPK2 sustains autophagic clearance are not yet elucidated but might involve the targeting of autophagy proteins still to be defined. Recently, several DAPK2 interacting partners or substrates have been identified in the autophagy protein network, including inhibitory interaction with 14-3-3, interference with mammalian target of rapamycin complex inhibition, or participation in the beclin interactome (50,51). Interestingly, DAPK2 silencing was also found to affect tumor necrosis factor–related apoptosis-inducing ligand signaling and nuclear factor-κB activation (52), reinforcing connections to cell inflammation. Clearly, more understanding is needed before a unified view on DAPK2 molecular action emerges, but in the context of obesity-related meta-inflammation, our present data identify the loss of DAPK2 expression and establish a link to attenuated autophagic clearance of adipocytes, revealing a potential novel actor in metabolic diseases.

Acknowledgments. The authors thank A. Kimchi (Weizmann Institute of Science, Rehovot, Israel) for providing bacterial clones containing human DAPK2 cDNA sequences and K52A kinase-dead mutant version, Xavier Le Liepvre and Francoise Lasnier (INSERM, Paris, France) for virus construction, Joan Tordjman (UMR_S 1166, Paris, France) for providing paraffin-embedded adipose tissue sections, and Jean-Luc Bouillot (Ambroise Paré Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France) for providing adipose tissue biopsies.

Funding. Financial support from Cardiovasculaire Obésité Rein Diabète (CORDDIM, Ile de France region) to H.S., from Clinical Research Program (PHRC 02076 on “adiposity signals”) to K.C., and from The French National Research Agency (ANR-14CE12-0017-02 LIPOCAMD) to I.D. is acknowledged.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.S., S.R., R.A., C.P., S.M., M.P., and C.R. performed experiments. H.S., K.C., and I.D. designed the study and wrote the manuscript. I.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.